IntelliSee

Customer stories

Case Study: 27J Schools Deploys AI Weapon Detection Across Four Colorado Campuses

See how 27J Schools deployed AI weapon detection across four Colorado campuses using existing security cameras and no new camera hardware.

IntelliSee12 min read
IntelliSee weapon detection bounding box on a person crossing a school bus parking lot, captured on an existing exterior campus camera
File #014 — K-12 Education
K-12 Education — Weapon Detection

27J brought AI weapon detection to four Colorado high school campuses using the cameras it already owned.

4Campuses Covered
0New Cameras Required
$0Capital Hardware
Aug 2026Live, Day One of Term
3rdApproach to Weapons Detection in Three Years
27J bought a walk-through weapons screening system in fall 2024 and scheduled it for a January 2025 launch. The district suspended the program in December 2024, before it ever went live, citing equipment that failed to deliver accurate detection and timely alerts. The system that launched in August 2026 runs on cameras the district already owned.
K-12 Education
Weapon Detection
Brighton, Colorado
33 Schools District-Wide
2026
Deployment Footprint

Four Campuses. Roughly Five Miles Apart.

School District 27J serves Brighton, Commerce City, Thornton, Henderson, and surrounding Adams and Weld County communities — 33 schools and more than 23,000 students. For the 2026-2027 school year the district extended AI weapon detection to all three comprehensive high schools plus its alternative campus, a footprint that spans about five miles north to south.

Every one of those campuses was already covered by district-owned exterior cameras watching parking lots, stadiums, ball fields, and building perimeters. None of them had a system capable of doing anything with what those cameras saw.

Brighton
Thornton
Henderson
4
1
3
2
N
2 km
Campuses Under Detection
1
Brighton High School
Comprehensive high school, Brighton. The district's oldest, dating to the early 20th century.
2
Prairie View High School
Comprehensive high school, Henderson. Southernmost campus in the deployment.
3
Riverdale Ridge High School
Comprehensive high school, Thornton. The district's newest, with an on-campus STEM and CTE center.
4
Innovations and Options
Alternative campus, Brighton. Roughly half a mile from Brighton High School.
Parking Lots
Stadiums
Ball Fields
Building Exteriors

Campus locations approximate. Coverage zones as described in the district's public program materials.

The Challenge

Screening Everyone Is Expensive. Watching Everything Is Impossible.

Districts weighing weapons detection generally choose between two models. The first screens people at entry points. The second monitors the environment continuously. Each fails in a different way, and 27J had already learned the first lesson the hard way.

Entry screening carries costs well beyond the hardware. It requires staffing at every screened door, every morning. It creates queues that push into first period. And it protects only the doors it covers, for only the minutes it is running — which leaves parking lots, athletic fields, and stadium areas outside the perimeter entirely.

The demand was not abstract. Access to weapons was among the concerns students themselves raised in a district safety survey, which the Student Advisory Council presented to the school board in May 2026. The district was responding to a problem its own students had named.

Before Deployment
A purchased walk-through screening program suspended before launch over accurate detection and timely alerts
Exterior cameras across four campuses recording, not detecting
No staff assigned to watch live camera feeds at any hour of the day
Parking lots, stadiums, and ball fields outside any screening perimeter
Students naming weapons access as a top concern in the district's own safety survey
Fall 2024
Screening purchasedWalk-through weapons screening announced for the high schools, slated for a January 2025 launch.
Dec 2024
Program suspendedSuspended before go-live. District cites failure to deliver accurate detection and timely alerts.
Mar 2026
IntelliSee acceptedThree-year agreement covering four campuses, funded through grant dollars.
May 2026
Students weigh inStudent Advisory Council presents safety survey results to the board. Weapons access is on the list.
Aug 2026
Detection liveRunning on existing cameras at all four campuses from the first day of the school year.
The Technical Gap

A Human Watching Live Video Stops Seeing It.

Continuous camera monitoring has the opposite problem from entry screening: coverage without attention. Research commissioned by the National Institute of Justice found that an operator watching live video begins missing a substantial share of screen activity after roughly twelve minutes, and misses the overwhelming majority of it after twenty-two. The effect compounds as the number of monitored feeds grows.

A district with hundreds of cameras and a handful of staff is not really monitoring hundreds of cameras. It is recording them. That distinction only becomes visible after an incident, when the footage is reviewed and the moment everyone missed is already hours old.

First minutes on shift
Low
After 12 minutes
45%
After 22 minutes
95%
Screen activity missedUp to 100%

Figures as reported in The Appropriate and Effective Use of Security Technologies in U.S. Schools, National Institute of Justice. Software does not fatigue on the same curve; it evaluates every frame the same way at minute one and minute four hundred.

Unfortunately, we don't have folks who watch the cameras and monitor them 24/7 like we would with the AI detection. In the current process, we don't have anything that would alert us that quickly.
Sam Ortega, Director of Safety and Security, 27J Schools — to CBS Colorado
The Paradigm Shift

Reactive Security vs. Proactive Safety

Before IntelliSee
Cameras record. Footage is reviewed after an incident is already over
Coverage limited to screened doors during screened hours
Parking lots, stadiums, and ball fields outside the perimeter entirely
Staffing cost at every covered entrance, every morning
The intervention window closes before anyone knows it opened
With IntelliSee
Every designated feed evaluated continuously, without fatigue
Coverage follows the cameras — lots, fields, stadiums, exteriors
Candidate detections verified across frames, then by a trained reviewer
Confirmed alerts routed to district safety staff and, via RapidSOS, to dispatch
No new cameras, no queues, no change to the school day
The Approach

Put the Attention Where the Cameras Already Are.

IntelliSee is camera-agnostic software rather than a hardware product. It analyzes video from cameras a facility already owns, which meant 27J could extend detection to its high school campuses without replacing infrastructure or adding fixtures outside its regular replacement cycle. The district stated publicly that families would see no new cameras and no change to the daily routine.

Weapon detection through computer vision is an object classification problem, not a facial one. The model is trained on the visual signature of a firearm — shape, proportion, orientation, and how the object is held relative to a human form — and evaluates each frame against that signature. It has no concept of who is holding the object, and no need for one.

Camera Zones Covered
Student & Staff Parking Lots
Stadium & Spectator Areas
Athletic & Ball Fields
Building Exteriors & Approaches
Deployed Detection Type
Visible Weapon Detection
Firearms visible in frame on any covered exterior camera, verified across sequential frames before escalation
Available on the Same Feeds
Unauthorized Access
People in restricted zones or on campus outside permitted hours
Loitering
Dwell time in entry, service, and perimeter areas
Vehicle Detection
After-hours vehicle activity in lots and service approaches
Rooftop Intrusion
Any person on roof surfaces or access ladders
Slip and Fall Risk
Spills and hazards in high-traffic corridors and entries
Smoke and Fire
Visual smoke and flame before a heat sensor would trip
From Frame to Response

A Detection Is Not an Alarm Until a Person Agrees With It.

Step 01
Existing Camera Feed
District-owned exterior cameras stream to the IntelliSee appliance. No new fixtures, no re-cabling.
Step 02
Object Model Evaluates the Frame
The model scores each frame against the visual signature of a firearm. Identity is never an input.
Step 03
Multi-Frame Confirmation
A shape that appears in one frame and vanishes in the next does not survive. Persistence separates detection from motion sensing.
Step 04
Trained Human Verification
Candidate detections route to IntelliSee reviewers. Nothing escalates on the model's word alone.
Step 05
District Safety and Dispatch
A confirmed weapon notifies designated district personnel and, through the RapidSOS integration, emergency dispatch.
The unresolved branchAn image a reviewer cannot resolve with confidence goes back to the district for additional review rather than out as a 911 call. From there the district's own protocols take over. 27J has been explicit in its public materials that the software is a detection and alerting tool — it does not replace human judgment, school safety staff, or the emergency procedures already in place.
Real IntelliSee platform output showing a weapon detection bounding box on a person walking through a school bus parking lot
Real platform output. A visible firearm classified on an existing exterior camera covering a school bus lot — the class of area that sits outside any entry-screening perimeter. Faces are obscured in published detection imagery; the platform does not perform facial recognition.
The Investment

What Four Campuses of Coverage Actually Cost.

Districts rarely get a straight answer on price, so here is the public record. 27J's agreement with IntelliSee is a three-year program covering four campuses, funded through grant dollars rather than the district's general fund.

3 YearsContract Term

Accepted March 2026. Live at all four campuses on the first day of the 2026-2027 school year.

$179,150Total Program Cost

Roughly $60,000 per year across four campuses. Cost scales with campus count and camera coverage, not enrollment.

$0Capital Hardware

No new cameras. No screening equipment. No scanning staff. No permanent bottleneck at the start of the school day.

The headline number is the easy part of a safety budget. The hard part is everything that sits underneath it — and that is where the two models separate.

Cost Line
Entry Screening
Software on Existing Cameras
Capital equipment
Scanners or turnstiles at every covered entrance
None
Recurring staffing
Operators at every screened door, every morning
None added
Impact on the school day
Queues at entry that push into first period
No change to the daily routine
Coverage beyond entrances
Lots, fields, and stadiums fall outside the perimeter
Any camera the platform can see
Hardware replacement cycle
Device refresh on the vendor's schedule
Follows the district's existing camera cycle
Adding a second threat type
New vendor, new contract, new hardware
Same feeds, same platform
Funding path
Often competes with general fund priorities
Fits school safety grant programs — how 27J funded it

Because there is no capital equipment purchase, the program fits cleanly into school safety grant funding rather than competing with facilities or instruction for general fund dollars. If you want to model the arithmetic for your own campus count and camera inventory, the Proactive Safety Value Calculator runs it against your numbers rather than someone else's.

Contract figures as reported by The Brighton Blade and CBS Colorado, August 2026.

Intelligence Brief

Privacy by Design: Why Object Recognition and Facial Recognition Are Different Problems

The first question most parents ask about AI on school cameras is whether the system is identifying their child. The honest answer is architectural rather than reassuring-sounding. IntelliSee does not use facial recognition. It does not collect or generate biometric identifiers — no faceprints, fingerprints, voiceprints, or retina scans. It does not maintain a database of identities or a watchlist, because identity is not an input the detection models use.

A weapon detection model is trained to answer one question: is there a firearm in this frame? Adding facial recognition would not improve that answer. It would introduce a category of data the system has no use for and every reason to avoid. 27J published this framing in its own words, stating that the system does not identify people, maintains no database or watchlist, and will not be used for investigative purposes or to discipline students for conduct unrelated to safety. The district also published a formal AI disclosure statement ahead of Colorado's forthcoming requirements for deployers of automated decision-making technology, including the right to ask whether an AI detection factored into a district decision, to request human review, and to request correction of inaccurate personal data.

What Other Districts Should Copy

The Communication Plan Was Part of the Deployment.

Most of the friction districts encounter when adopting safety technology is not technical. It is informational. Families hear that AI is watching their children and fill the vacuum with the worst available interpretation. 27J closed that vacuum before it opened.

Ahead of the first day of school, the district published four things on its own website:

  • A program overview page explaining in plain language what the software is, what it does, and what it does not do.
  • A written FAQ answering the questions the district anticipated most — from whether new cameras were coming to what happens in the event of a false alarm.
  • A formal AI disclosure statement with the scope of the system and the rights available to students and families.
  • A letter home from each high school principal, signed and specific to that campus.

“We understand the use of this new technology may create questions for students and families,” the Brighton High School letter opened. That sentence does more work than it appears to. It concedes the concern before anyone has to raise it, which is the difference between a district that is communicating and a district that is defending.

IntelliSee has since packaged this approach into a launch communications kit — editable parent and staff letter templates, a district-ready FAQ, suggested signage language, and a press kit — so districts deploying after 27J do not have to build the messaging from a blank page. Contact us to request it.

In the News

Coverage of the 27J Deployment

Members of the media can find brand assets, executive bios, and contact details in the IntelliSee Media Room.

Common Questions

What districts ask before they deploy.

Talk to an Education Specialist →

No. IntelliSee is camera-agnostic software that analyzes video from the cameras a district already owns. 27J extended detection to four campuses without installing new cameras outside its regular replacement schedule. For most districts the practical constraint is camera placement and field of view, not camera brand.

Cost scales with campus count and camera coverage rather than enrollment. Publicly reported figures put 27J's three-year agreement at $179,150, roughly $60,000 per year across four campuses, funded through grant dollars rather than the general fund. Because the platform runs on existing cameras there is no capital hardware purchase and no per-door staffing cost. Run the numbers for your own footprint with the Proactive Safety Value Calculator.

IntelliSee does not use facial recognition, does not collect or generate biometric identifiers, and does not maintain a database of identities or a watchlist. Weapon detection is an object classification problem. Identity is not an input the models use, and adding it would not improve detection.

Publish before you launch. Districts that put out a plain-language program page, a written FAQ, an AI disclosure statement, and a signed letter from each building principal tend to face far fewer questions after go-live. 27J published all four ahead of the first day of school. IntelliSee provides editable templates for each to districts deploying the platform.

Candidate detections are verified across sequential video frames and then reviewed by trained IntelliSee personnel before anything escalates. An image a reviewer cannot resolve with confidence is returned to the district for additional review rather than sent out as a 911 call. The district's own emergency protocols govern every step after notification.

People are still making the decisions — our response teams, our safety and security team, and our local law enforcement.
Sam Ortega, Director of Safety and Security, 27J Schools — to CBS Colorado
Get Started

Your Cameras Already See
The Parking Lot.
Is Anything Watching?

A risk assessment maps your existing camera coverage against the threats you are actually trying to prevent, and shows which of them are detectable on the infrastructure you already own.

Your people. Your spaces.

Put a clearer view to work.

See how IntelliSee can bring useful context from your compatible cameras to the people who need it.